A personal agent that builds programs for my life and runs them with me.
I talk to it in chat. It builds programs I actually keep using — not throwaway demos — and plugs into the tools I already live in.
Want to try it out for yourself? Message meWhy I call them programs
I’ve built a hundred throwaway apps on ChatGPT, Claude, and Lovable — most of which I used once, at the moment I made them.
Regent builds programs I use every day. They’re made for an agent first: Regent pilots them, I can run the whole thing through chat, and they still have a real UI when I want one. Fully tested, same backend as Regent — not a demo that dies after the screenshot.
No app deploys. Programs hot-reload as they change. I customize them as I go — just by chatting. I open them on the web, as a PWA on my phone, or as a Telegram Mini App. Same program, same state, no shipping step.
How it works for me
01
I ask
I tell Regent what I need — a tracker, a list, something that keeps my life sorted.
02
It builds a program
A subagent ships a real program for me, fully tested. No deploy — it just shows up.
03
One backend
My programs and the agent share the same backend — memory, data, and tools stay together.
04
Keep shaping it
I change a program over chat; it hot-reloads. Regent pilots it, or I open the web / PWA / Telegram UI.
Programs are agent-ready and fully tested before I rely on them. No separate deploys — they run on the shared backend Regent already uses for me.
What I connect
It reaches into the places I already am — Google Workspace, WhatsApp, X — so it can help me stay organized and actually get things done.
- GmailGoogle Workspace
- CalendarGoogle Workspace
- SheetsGoogle Workspace
- X / Twitter
Programs it built for me
A few of the programs it’s built and still runs for me — on the web, as a PWA, and as Telegram Mini Apps.
Common questions
- What is a Regent program?
- A program is a personal app Regent builds for me: agent-ready, fully tested, sharing Regent’s backend. No separate deploy — it hot-reloads, I customize it over chat, and I use it on the web, as a PWA, or as a Telegram Mini App.
- How is this different from apps from ChatGPT, Claude, or Lovable?
- I’ve built a hundred throwaway apps on ChatGPT, Claude, and Lovable — most of which I used once, at the moment I made them. Regent builds programs I use every day. They’re made for an agent first: Regent pilots them, I can run the whole thing through chat, and they still have a real UI when I want one. Fully tested, same backend as Regent — not a demo that dies after the screenshot.
- Do Regent programs need to be deployed?
- No app deploys. Programs hot-reload as they change. I customize them as I go — just by chatting. I open them on the web, as a PWA on my phone, or as a Telegram Mini App. Same program, same state, no shipping step.
- How does Regent work day to day?
- I ask: I tell Regent what I need — a tracker, a list, something that keeps my life sorted. It builds a program: A subagent ships a real program for me, fully tested. No deploy — it just shows up. One backend: My programs and the agent share the same backend — memory, data, and tools stay together. Keep shaping it: I change a program over chat; it hot-reloads. Regent pilots it, or I open the web / PWA / Telegram UI.